|
Hui Ren (任辉) I am a first-year UIUC CS Ph.D. student, advised by Prof. Alexander Schwing. I hold a Bachelor’s degree in Computer Science at ShanghaiTech University, and I am an exchange student at MIT in 2024 Spring. I was very fortunate to join the MIT CSAIL Vision Group advised by Prof. Antonio Torralba, Prof. David Bau and mentor Dr. Joanna Materzyńska. I was also a visiting student at UCLA Visual Machines Group advised by Prof. Achuta Kadambi and mentor Mr. Shijie Zhou. From Feb. 2023 to Feb. 2024, I acted as an undergraduate research assistant in PLUS Lab, advised by Prof. Xuming He and mentor Dr. Chuyu Zhang.
|
|
ResearchI am interested in representation learning and its application in semantic and related scenarios. My current research work focuses on Generative AI and 3D Vision. Previously, I explored Optimal Transport theory for Unsupervised Deep Clustering and 3D Semi-Supervised Learning, with a focus on imbalanced and open-world scenarios. Generally speaking, my research aim is advancing intelligent systems that can simulate, understand, and interact with the world while efficiently leveraging prior knowledge to generalize across diverse real-world scenarios. - How far are we from the genuine advent of artificial intelligence? I am always open to collaboration and discussion. If you are interested in my research or have any questions, please feel free to reach out! |
News
|
PublicationsNote: * indicates equal contribution. |
Selected course projects |
|
Business War Policy Exploration
Artificial Intelligence We proposed a business war model in which restaurants must balance attracting customers and making profits to maximize revenue and even defeat competitors. We implemented and compared various AI strategies, including Expectimax search, Neural-Network prediction, and Reinforcement learning algorithms to explore the optimal policy. Report / Code |
|
Ray tracing based multi-resolution iso-surface rendering
Computer Graphics I We rendered the iso-surface of vortices in a multi-resolution fluid velocity field based on ExaBricks data structure. We implemented a bitmap accelerated KD-tree and BVH for multi-resolution data to support fast ray marching. We proposed an advanced adaptive sampling strategy to solve the problem of rendering surface discontinuity, improving the rendering speed by more than 50x at the same precision. We created a UI for conveniently adjusting the parameters and rendering. Report / Code |
Activities
|
Selected Awards and Honors
|
This homepage is designed based on Jon Barron's website.
© 2023-2024 Hui Ren